Go is a statically typed, compiled programming language originally designed at Google. Go supports arm64 out of the box, and available in all common distributions, with recent changes that improve performance, so make sure to use the latest version of the Go compiler and toolchain.
The main implementation of the Go compiler, golang/go, has improved performance on Arm by implementing a new way of passing function arguments and results using registers instead of the stack. This change has been available on x86-64 since 1.17, where it brought performance improvements of about 5%. On Arm this change typically gives even higher performance improvements of 10% or more.
To learn more about the use cases benefiting from Go 1.18's performance improvements, check the blog post: Making your Go workloads up to 20% faster with Go 1.18 and AWS Graviton.
The main implementation of the Go compiler, golang/go, has improved performance for the following standard library packages:
- crypto/ed25519 - the package has been rewritten, and all operations are now approximately twice as fast on both arm64 and amd64.
- crypto/elliptic - CurveParams methods now automatically invoke faster and safer dedicated implementations for known curves (P-224, P-256, and P-521) when available. The P521 curve implementation has also been rewritten and is now constant-time and three times faster on amd64 and arm64.
The main implementation of the Go compiler, golang/go, has improved performance on Arm with couple of changes listed below. Building your project with Go 1.16 will give you these improvements:
- ARMv8.1-A Atomics instructions, which dramatically improve mutex fairness and speed on Graviton 2, and modern Arm core with v8.1 and newer instruction set.
- copy performance improvements, especially when the addresses are unaligned.
Changes to commonly used packages that improve performance on Arm can make a noticeable difference in some cases. Here is a partial list of packages to be aware of.
| Package | Version | Improvements |
|---|---|---|
| Snappy | as of commit 196ae77 | assembly implementations of the hot path functions were ported from amd64 to arm64 |
Go automatically assigns a sensible value to GOMAXPROCS based on the number of
CPU cores available. However, using a container with a limitation on how much
CPU is available to that container can lead to problems. For example, using the
CFS scheduler
option in Docker, --cpus=1 can limit the available CPU time to the
equivalent of 1 CPU while still exposing all of the actually available CPUs to
the container. If you use CPU limits in this way, it may make sense to also
manually set GOMAXPROCS to an equivalent value.